Data Collection and Preprocessing¶
In [1]:
import pandas as pd
df = pd.read_csv('covid_19_data.csv',parse_dates=['Last Update'],index_col=0)
#parsing according to the last update datetime column
#serial number as index
df.rename(columns={'ObservationDate':'Date', 'Country/Region':'Country'}, inplace=True)
#renaming certain columns
In [2]:
#used for Map later
df_confirmed = pd.read_csv("time_series_covid_19_confirmed.csv")
df_confirmed.rename(columns={'Country/Region':'Country'}, inplace=True)
df_confirmed = df_confirmed[["Province/State","Lat","Long","Country"]]
df_temp = df.copy()
df_temp['Country'].replace({'Mainland China': 'China'}, inplace=True)
df_latlong = pd.merge(df_temp, df_confirmed, on=["Country", "Province/State"])
Exploratory Data Analysis (EDA)¶
In [3]:
print("Rows : ",df.shape[0])
print("\nColumns : ",df.shape[1])
print("\nFeatures : ",df.columns.tolist())
print("\nMissing Values : ",df.isnull().sum().sum())
print("\nUnique values : \n", df.nunique())
In [4]:
df.info() #information about the datatypes
In [5]:
print("Basic Statistics : \n",df.describe()) #metrics
In [6]:
print("Earliest Cases : \n",df.head())
In [7]:
print("Latest Cases : \n",df.tail())
In [8]:
print("\n\t\t Date-wise number of cases in each category\n")
df.groupby('Date').sum()
#total number of Confirmed cases , Deaths and Recovery per day.
Out[8]:
In [9]:
print("\n\t\tMaximum number of Confirmed,Deaths and Recovered Cases\n")
df1 = df.groupby(['Country', 'Province/State'])['Confirmed', 'Deaths', 'Recovered'].max()
df1 = df.groupby('Date')['Confirmed', 'Deaths', 'Recovered'].sum().reset_index()
df1 = df1[df1['Date']==max(df1['Date'])].reset_index(drop=True)
df1.style.background_gradient(cmap='Pastel1')
Out[9]:
In [10]:
print("\n\t\tWorld View - Country wise\n")
df_grouped = df.groupby('Country')['Confirmed', 'Deaths', 'Recovered'].sum().reset_index()
df2 = df_grouped.sort_values(by='Confirmed', ascending=False)
df2 = df2.reset_index(drop=True)
df2.style.background_gradient(cmap='summer')
#Maximum number of cases in China followed by Italy and Iran
Out[10]:
In [11]:
print("\n\t\tChina \n")
China = df.query('Country=="Mainland China"').groupby("Last Update")[['Confirmed', 'Deaths', 'Recovered']].sum().reset_index()
China
#Exploring data of china
Out[11]:
In [12]:
print("\n\t\tCountry Wise - Sorted(Alphabetically) order\n")
df.groupby("Country")[['Confirmed', 'Deaths', 'Recovered']].sum().reset_index()
Out[12]:
Visualisation¶
In [13]:
confirmed = df.groupby('Date').sum()['Confirmed'].reset_index()
deaths = df.groupby('Date').sum()['Deaths'].reset_index()
recovered = df.groupby('Date').sum()['Recovered'].reset_index()
In [14]:
import plotly.graph_objects as go
fig = go.Figure()
fig.add_trace(go.Bar(x=confirmed['Date'],
y=confirmed['Confirmed'],
name='Confirmed',
marker_color='blue'
))
fig.add_trace(go.Bar(x=deaths['Date'],
y=deaths['Deaths'],
name='Deaths',
marker_color='Red'
))
fig.add_trace(go.Bar(x=recovered['Date'],
y=recovered['Recovered'],
name='Recovered',
marker_color='Green'
))
fig.update_layout(title='Worldwide Corona Virus Cases - Confirmed, Deaths, Recovered (Bar Chart)',
xaxis_tickfont_size=14,
yaxis=dict(
title='Number of Cases',
titlefont_size=16,
tickfont_size=14,
),
legend=dict(
x=0,
y=1.0,
bgcolor='rgba(255, 255, 255, 0)',
bordercolor='rgba(255, 255, 255, 0)'
),
barmode='group',
bargap=0.15, # gap between bars of adjacent location coordinates.
bargroupgap=0.1 # gap between bars of the same location coordinate.
)
fig.show()
#steep rise in confirmed cases around 12th March 2020. Reason could be more tests and results
#available
In [15]:
n = 5
c_lrgst = df_grouped.Confirmed.nlargest(n)
d_lrgst = df_grouped.Deaths.nlargest(n)
r_lrgst = df.Recovered.nlargest(n)
top_5 = df_grouped.query('Confirmed in @c_lrgst')
In [16]:
fig = go.Figure()
fig.add_trace(go.Scatter(x=top_5['Country'],
y=top_5['Confirmed'],
mode='lines+markers',
name='Confirmed',
line=dict(color='blue', width=2)
))
fig.add_trace(go.Scatter(x=top_5['Country'],
y=top_5['Deaths'],
mode='lines+markers',
name='Deaths',
line=dict(color='Red', width=2)
))
fig.add_trace(go.Scatter(x=top_5['Country'],
y=top_5['Recovered'],
mode='lines+markers',
name='Recovered',
line=dict(color='Green', width=2)
))
fig.update_layout(
title='Top 5 Countries with Corona Virus Cases (Line Chart)',
xaxis_tickfont_size=14,
yaxis=dict(
title='Number of Cases',
titlefont_size=16,
tickfont_size=14,
),
legend=dict(
x=0,
y=1.0,
bgcolor='rgba(255, 255, 255, 0)',
bordercolor='rgba(255, 255, 255, 0)'
)
)
fig.show()
#China cases far more than the other top 5 countries
In [17]:
import plotly.graph_objects as go
import numpy as np
z = np.random.poisson(size=(len(China['Confirmed']), len(China['Last Update'])))
fig = go.Figure(data=go.Heatmap(
z=z,
x=China['Last Update'],
y=China['Confirmed'],
colorscale='fall'))
fig.update_layout(
title='China - Number of Cases per day',
xaxis_nticks=5)
fig.show()
#Cases are still increasing . No notion of decrease till now.
In [18]:
import plotly.express as px
fig = px.density_mapbox(df_latlong,
lat="Lat",
lon="Long",
hover_name="Province/State",
hover_data=["Confirmed","Deaths","Recovered"],
animation_frame="Date",
color_continuous_scale="Portland",
radius=7,
zoom=1,height=700)
fig.update_layout(title='Worldwide Corona Virus Cases Time Lapse - Confirmed, Deaths, Recovered',
font=dict(family="Courier New, monospace",
size=14,
color="#7f7f7f")
)
fig.update_layout(mapbox_style="open-street-map", mapbox_center_lon=0)
fig.update_layout(margin={"r":0,"t":0,"l":0,"b":0}) #tight layout - right , top, left , below
fig.show()
#trend of spread and at what rate. Huge jump in March
Forecasting / Prediction¶
Prophet Algorithm¶
Worldwide - Confirmed Cases¶
In [19]:
from fbprophet import Prophet
In [20]:
#Fixing data for forecasting
confirmed = df.groupby('Date').sum()['Confirmed'].reset_index()
confirmed.columns = ['ds','y']
confirmed['ds'] = pd.to_datetime(confirmed['ds'])
confirmed['cap']= 10000000
confirmed.head()
Out[20]:
In [21]:
m = Prophet(interval_width=0.95,yearly_seasonality=True,daily_seasonality=True,growth='logistic')
m.fit(confirmed)
future = m.make_future_dataframe(periods=61)
future.tail()
Out[21]:
In [22]:
future['cap']=10000000
forecast = m.predict(future)
forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail()
Out[22]:
In [23]:
from fbprophet.plot import plot_plotly
import plotly.offline as py
py.init_notebook_mode()
fig = plot_plotly(m, forecast) # This returns a plotly Figure
py.iplot(fig)
In [24]:
#help(Prophet)
In [25]:
m.plot_components(forecast);
Worldwide - Deaths¶
In [26]:
#fixing data
deaths.columns = ['ds','y']
deaths['ds'] = pd.to_datetime(deaths['ds'])
In [27]:
m = Prophet(interval_width=0.95,yearly_seasonality=True,daily_seasonality=True)
m.fit(deaths)
future = m.make_future_dataframe(periods=61)
future.tail()
Out[27]:
In [28]:
forecast = m.predict(future)
forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail()
Out[28]:
In [29]:
fig = plot_plotly(m, forecast) # This returns a plotly Figure
py.iplot(fig)
In [30]:
m.plot_components(forecast);
Worldwide - Recovery¶
In [31]:
#fixing data
recovered.columns = ['ds','y']
recovered['ds'] = pd.to_datetime(recovered['ds'])
In [32]:
m = Prophet(interval_width=0.95,yearly_seasonality=True,daily_seasonality=True)
m.fit(recovered)
future = m.make_future_dataframe(periods=61)
future.tail()
Out[32]:
In [33]:
forecast = m.predict(future)
forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail()
Out[33]:
In [34]:
fig = plot_plotly(m, forecast) # This returns a plotly Figure
py.iplot(fig)
In [35]:
m.plot_components(forecast);
In [ ]: